This study focuses on three questions relevant to in vivo gene transfer with replication deficient recombinant adenovirus (Ad) vectors. (1) Can Ad vectors deliver a gene coding for a secreted protein sufficient for therapeutic delivery of the protein? (2) Can immunosuppressive therapy suppress the anti-Ad humoral response sufficiently to permit effective gene transfer upon repetitive administration of the same vector? (3) Can an anti-Ad humoral response be circumvented using an Ad vector of an alternative serotype? The clinical model to evaluate these biologic questions is the thrombocytopenia associated with high dose chemotherapy for stage IV breast carcinoma. The gene to be delivered is human thrombopoetin cDNA, a growth factor for megakaryocyte progenitors. The protocol utilizes two E1a , partial E1b , partial E3 Ad vectors carrying the TPO cDNA, one based on Ad serotype 5 (Ad5CMV.TPO) and one based on Ad serotype 2 (Ad2CMV.TPO). Treatment of thrombocytopenia was chosen because the clinical model requires: (1) sufficient expression to deliver a protein systemically (permitting evaluation of the robustness of Ad- mediated in vivo gene transfer); (2) TPO cDNA expression for only 1 to 2 weeks (eliminating the problem of cellular immune recognition of viral antigens and elimination of the cells expressing the TPO cDNA); and (3) repetitive administration of the vector in the context of chemotherapy for the underlying malignancy (permitting evaluation of immunosuppression on anti-Ad humoral response, as well as the evaluation of the ability of administration of an Ad vector of an alternative serotype).